{"slug": "recopii-a-relation-grounded-framework-for-building-contextual-pii-masking", "title": "ReCoPII: A Relation-Grounded Framework for Building Contextual PII Masking Datasets", "summary": "A research team introduced ReCoPII, a relation-grounded framework that generates contextual PII masking datasets without relying on real personal data, producing the ReCoPII-Ko corpus from Korean documents across six domains. Across 300 pairs, four annotators agreed with 88.42% of ReCoPII's sampled KEEP labels, versus 26.27% for SPY-style augmentation of the same sources. Classifiers trained on the corpus over-mask less than the SPY reference on a matched test, and on human-verified entity records the four detectors meeting a fixed leakage budget over-mask at most 5.04% with at most 4.18% leakage, while the detector performs comparably to XLM-R-large on KDPII with about a quarter of its parameters.", "body_md": "arXiv:2610.35925v1 Announce Type: new \nAbstract: Training contextual masking systems for personally identifiable information (PII) requires examples distinguishing maskable from preservable mentions of one entity type, but privacy restrictions limit access to real data. Author-based augmentation (SPY) labels mentions unrelated to the author as non-PII, presuming a single data subject and overlooking others. We propose ReCoPII (Relation-grounded Contextual PII masking), a framework that builds such data without this premise: it plans ontology-constrained entity relations, writes insertion sentences with placeholders, reconstructs annotations with code, checks the result with a separate LLM, and derives MASK/KEEP labels from entity relations. On Korean documents from six domains, it yields ReCoPII-Ko, a corpus with relation, privacy, and MASK/KEEP annotations, which we validate. Across 300 pairs, four annotators agree with 88.42% of its sampled KEEP labels, against 26.27% for SPY-style augmentation of the same sources. Classifiers trained on it over-mask less than under that reference on a matched test. Contextual masking fine-tuned jointly with a PII detector removes most over-masking of detection alone within a fixed leakage budget; on human-verified entity records, the four detectors meeting it over-mask at most 5.04% with at most 4.18% leakage. The detector is comparable to XLM-R-large on KDPII with about a quarter of its parameters.", "url": "https://wpnews.pro/news/recopii-a-relation-grounded-framework-for-building-contextual-pii-masking", "canonical_source": "https://arxiv.org/abs/2610.35925", "published_at": "2026-10-28 04:00:00+00:00", "updated_at": "2026-10-28 04:12:26.293556+00:00", "lang": "en", "topics": ["artificial-intelligence", "machine-learning", "natural-language-processing", "ai-research", "ai-safety"], "entities": ["ReCoPII", "ReCoPII-Ko", "SPY", "XLM-R-large", "KDPII"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/recopii-a-relation-grounded-framework-for-building-contextual-pii-masking", "markdown": "https://wpnews.pro/news/recopii-a-relation-grounded-framework-for-building-contextual-pii-masking.md", "text": "https://wpnews.pro/news/recopii-a-relation-grounded-framework-for-building-contextual-pii-masking.txt", "jsonld": "https://wpnews.pro/news/recopii-a-relation-grounded-framework-for-building-contextual-pii-masking.jsonld"}}